Getty Images

Rubrik pivots to generative AI with Predibase acquisition

Data stored within Rubrik's backup platform could soon be another data lake for customers to tap into as the vendor looks to acquire AI specialists Predibase.

Rubrik is making another pivot to creating data lakes for generative AI through a new acquisition.

Rubrik, a SaaS data backup platform vendor, announced last week that it has agreed to acquire AI specialist Predibase. The acquisition will bring GenAI capabilities to data stored within Rubrik's platform, according to Rubrik spokespeople.

Successful AI applications in the enterprise rely on large data stores, said Christophe Bertrand, an analyst at TheCube Research. Backup platforms already possess much of an enterprise's data, so bringing that data closer to AI would expedite projects.

"They're really positioning themselves as a source of data," Bertrand said. "This acquisition is all about accelerating ROI for AI for their clients."

Rubrik declined to answer questions from Informa TechTarget regarding acquisition costs, operations and other details. Other media outlets like CNBC reported that Rubrik intends to pay between $100 million to $500 million for Predibase and said the company will operate as a separate unit after the deal closes.

Next pivot

Predibase sells tools for fine-tuning what are known as small language models (SLMs). These are derivatives of large language models (LLMs), such as the ubiquitous ChatGPT or Google Gemini, for specific uses.

The vendor's platform enables customers to refine open source models like Meta's Llama or those from Mistral AI into customer SLMs for specific workloads and applications, according to Predibase.

SLMs are likely the future of AI within many enterprises, said Scott Hebner, an analyst at TheCube Research. LLMs introduce outside knowledge and security risks and complicate AI deployments compared with the more refined knowledge base and resolves that SLMs could provide, he said.

"The days of businesses using general-purpose, publicly available AI are gone," Hebner said.

In a blog post detailing the acquisition, Rubrik CEO Bipul Sinha wrote that the Rubrik platform created data lakes, Bertrand noted. By saying Rubrik now supports data lake creation, the vendor has announced another industry pivot after recently dipping its toes into cybersecurity.

They're still a backup and recovery company, let's not kid ourselves.
Christophe BertrandAnalyst, TheCube Research

Rubrik saying it helps create data lakes versus implementing data lake functions is very different, he said. Such a motivation will require the vendor to establish partnerships and connectivity with a whole new technology coterie.

"They're still a backup and recovery company, let's not kid ourselves," Bertrand said. "[But] how do you execute on [AI] when you don't come from that space?"

Rubrik is the first backup vendor to explicitly discuss using data within backups for LLM or SLM implementations, said Jerome Wendt, CEO and founder of Data Center Intelligence Group.

Cohesity's Gaia generative AI service can query backups within its platform, but isn't currently exporting data for other applications, he said.

The other major challenge facing Rubrik will be the customers themselves, according to Wendt. Enterprises have spent the past few years vetting and analyzing AI, but many have found that they simply lack enough accurate and relevant data to build AI data lakes and agentic workflows, he said.

"The biggest problem for organizations is going to be the quality of your data," Wendt said. "Many organizations are not dealing with exabytes or petabytes [of files], but they still have to go through that data and ingest it."

Tim McCarthy is a news writer for Informa TechTarget covering cloud and data storage.

Dig Deeper on Data backup and recovery software